Advent of the Internet

If we talk to an engineer responsible for systems infrastructure, we’ll eventually hear the word “density.” Hardware and data center floor space are expensive. The goal of every data center operations manager is to pack as many services into the least amount of hardware possible. A single server costs the same whether it supports one client or one hundred. Higher density—more websites and web services sharing the same hardware—leads to lower operating costs and greater profit margins.

The story of containers in general, and of Docker in particular, is a story of density and scale. The best way to understand that story is to learn a little about the history that led to the development of containers.

The beginning of the internet#

In the early days of the internet, when Yahoo! and AOL were giants, and Google wasn’t even an idea yet, the concept of a server was clear and well-defined. A server was a physical computer, in many ways similar to a desktop, but with greater power and reliability. Servers were self-contained, each with its own CPU, memory, storage, and networking components. Most importantly, each had an operating system: one operating system. This one-to-one relationship between servers and operating systems made it expensive to scale up to increase capacity. When the capacity limit of a server was reached, the only option was to add more servers (also known as horizontal scaling) at a cost of many thousands of dollars each. The hardware and software costs to scale up a website grew linearly with demand.